Dumb question about headless & Nebraska

Lex Spoon lex at cc.gatech.edu
Wed Apr 4 23:20:34 UTC 2001


Bert Freudenberg <bert at isg.cs.uni-magdeburg.de> wrote:
> On Mon, 2 Apr 2001, Stephen Pair wrote:
> 
> > Yes...but, most Windows programs (including the Win32 squeak vm) do not
> > output their display to an X server, they use GDI.  So, unless you have a
> > replacement for the windows GDI that converts GDI calls to the X11 protocol,
> > it doesn't do you much good on the windows platform (for remotely accessing
> > windows programs).  If you're looking for a cheap X server on windows, Xvnc
> > might be an option...if you're looking to remotely access an instance of
> > Squeak running on Windows, that's different.
> 
> Nah, Squeak uses only one drawing primitive, and non of fancy GDI
> functions. It should be not that hard to rip the X displaying code from
> the Unix sources and put it into the Win32 code. There are X apps for
> Windows (comes with Hummingbird Exceed).
> 

It depends at which level you look.  The image<->VM interface only uses
ony function, but in Morphic, drawing is done on Canvas's using commands
like "draw a line" and "draw text with this font".  Unless you are using
bitmaps, then the "draw a line" style commands are more efficient than
the single forceToScreen() primitive that is used between image and VM. 
(On the other hand, the bits have to be put on the display eventually,
and so forceToScreen() works just fine on local machines).

To get to the point, why not use Nebraska?!  It's pretty zippy for
working with text windows, and it's had very little work put into
optimization.  I'd think it's better to start with Nebraska and optimize
it, than to start with a model that doesn't fit morphic very well.

The only thing that bugs me is that nobody every implemented a password
system...  For that sole reason, I don't tend to leave it running,
myself.


-Lex





More information about the Squeak-dev mailing list